fatal internal error at ISIF=3 Structure Optimization

[okuno]

Dear Vasp Users.

I'm bigining to calculate the structure optimization with ISIF=3.
I found in the calculation process, the error massage like

curvature: 0.00 expect dE= 0.174-311 dE for cont linesearch 0.166-311
ZBRENT: fatal internal in brackting
system-shutdown; contact gK immediately

and the calculation stopped before it reach the conversence results.
When such error occurs, how should I treat it?

My INCAR file is below, it is antiferromagnetic insulater materials.

ENCUT=520
PREC=High
ISMEAR=-5
SIGMA=0.01
ISPIN=2
ICHARG=2
VOSKOWN=1
EDIFF=1e-5
MAGMOM=0 0 -3.0 3.0 0 0 0 0 0 0

AMIX=0.2
BMIX=0.00001
AMIX_MAG=0.8
BMIX_MAG=0.00001

NELMIN=6
ISIF=3
IBRION=2
NSW=100
EDIFFG=-0.001


LDAU=.TRUE.
LDAUL= -1 2 -1
LDAUU= 0.0 4.0 0.0
LDAUJ= 0.0 1.0 0.0
LDAUTYPE = 2
LDAUPRINT = 2

LWAVE=.FALSE.

[admin]

it seems that the equilibrium has been reached and the BRENT algorithm cannot optimize the geometry further (please note that EDIFFG=-0.001 is a very strict criterium!).
please
1) have a look if there are any further changes in the geometries (XDATCAR) and in the total energies of the last steps.
2) for very small corrections of the atomic positions (close to the minimum), the CG algorithm may fail. In case you want to relax further, please set
IBRION=1
ADDGRID=true
in INCAR